Method for increasing signal-to-noise ratio in magnetic resonance imaging using per-voxel noise covariance regularization

Patent metadata

Field	Value
Publication number	US-10429475-B2
Application number	US-201414774551-A
Country	US
Kind code	B2
Filing date	Mar 12, 2014
Priority date	Mar 12, 2013
Publication date	Oct 1, 2019
Grant date	Oct 1, 2019

Abstract

A method for maximizing the signal-to-noise ratio (“SNR”) in a combined image produced using a parallel magnetic resonance imaging (“MRI”) technique is provided. The image combination used in such techniques require an accurate estimate of the noise covariance. Typically, the thermal noise covariance matrix is used as this estimate; however, in several applications, including accelerated parallel imaging and functional MRI, the noise covariance across the coil channels differs substantially from the thermal noise covariance. By combining the individual channels with more accurate estimates of the channel noise covariance, SNR in the combined data is significantly increased. This improved combination employs a regularization of noise covariance on a per-voxel basis.

First claim

The invention claimed is:

1. A method for producing an image of a subject using a magnetic resonance imaging (MRI) system, the steps of the method comprising:
a) acquiring k-space data from the subject using an MRI system that includes an array of radio frequency (RF) receiver coils;
b) reconstructing an image for each receiver coil from the k-space data acquired in step a) from that receiver coil;
c) estimating a noise covariance matrix for each voxel in each image reconstructed in step b);
d) producing an image of the subject by weighting and combining the images reconstructed in step b) while regularizing the noise covariance matrices on a per-voxel basis.

2. The method as recited in claim 1 in which step d) includes regularizing the noise covariance matrices by performing a truncated singular value decomposition of the noise covariance matrices.

3. The method as recited in claim 2 in which step d) includes multiplying signals in the images reconstructed in step b) with a subset of singular values computed in the truncated singular value decomposition of the noise covariance matrices.

4. The method as recited in claim 3 in which the subset of singular values includes a selected number of the largest computed singular values.

5. The method as recited in claim 2 in which the truncated singular value decomposition produces a truncated matrix of singular values having a specified condition number.

6. The method as recited in claim 5 in which the specified condition number is selected to control the regularization of the noise covariance matrices.

7. The method as recited in claim 6 in which step d) includes determining a number of singular vectors that results in the specified condition number.

8. The method as recited in claim 1 in which the noise covariance matrices estimated in step c) are image-domain noise covariance matrices.

9. The method as recited in claim 8 in which the k-space data acquired in step a) is accelerated k-space data that is acquired by undersampling k-space, and in which step c) includes estimating the Image-domain noise covariance matrices from accelerated k-space data.

10. The method as recited in claim 9 further comprising acquiring pre-scan k-space data in which RF excitation is disabled, and in which step c) includes analytically deriving the image-domain noise covariance matrix for each voxel in each image reconstructed in step b) from the pre-scan k-space data.

11. The method as recited in claim 1 in which the k-space data acquired in step a) is representative of a time-series of images, and in which the noise covariance matrices estimated in step c) are time-series noise covariance matrices.

12. The method as recited in claim 11 further comprising acquiring pre-scan k-space data and reconstructing a time series of Images from the acquired pre-scan k-space data; and in which step c) includes estimating the time-series covariance matrices from the time series of images reconstructed from the acquired pre-scan k-space data.

13. The method as recited in claim 1 in which the noise covariance matrices estimated in step c) are noise covariance matrices that include combined effects of image-domain noise and time-series noise.

14. The method as recited in claim 1 in which step d) includes regularizing the noise covariance matrices by performing a Tikhonov regularization, and in which the performing the Tikhonov regularization includes selecting a regularization parameter.

15. The method as recited in claim 14 in which the regularization parameter is selected by empirically determining an optimal balance between artifact suppression and signal-to-noise ratio enhancement in the image of the subject.

16. The method as recited in claim 14 in which the regularization parameter is selected using an automatic technique for seeking a regularization parameter that achieves a balance between a data consistency term and a smoothness constraint term of a cost function.

17. The method as recited in claim 16 in which the automatic technique used to select the regularization parameter is an L-curve method.

Assignees

• Polimeni Jonathan R
• Setsompop Kawin
• Wald Lawrence L
• Massachusetts Gen Hospital

Inventors

• Polimeni Jonathan R
• Setsompop Kawin
• Wald Lawrence L

Classifications

• G01R33/385

  using gradient magnetic field coils · CPC title

• G01R33/5608 Primary

  Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels (image data processing or generation, in general G06T) · CPC title

• G01R33/5611

  Parallel magnetic resonance imaging, e.g. sensitivity encoding [SENSE], simultaneous acquisition of spatial harmonics [SMASH], unaliasing by Fourier encoding of the overlaps using the temporal dimension [UNFOLD], k-t-broad-use linear acquisition speed-up technique [k-t-BLAST], k-t-SENSE (structural details of arrays of sub-coils G01R33/3415) · CPC title

• G01V3/14 Primary

  operating with electron or nuclear magnetic resonance · CPC title